A Novel Iris Segmentation Approach Using Spindle Ternary Tree
Yu‐Hui Lin, Zubin Ye, Baosen Xiao · 2022
This paper presents an accurate and novel iris segmentation approach. First rough iris region is extracted by boosted Local Binary Pattern features classifier. Follow Intensity gradient matrix is then built though gray gradient operator under polar coordinate. Then Special Ternary Tree model is proposed to handle the matrix for tracking iris boundaries by searching the best route as iris boundary. Comparatively, experimental results on the popular iris databases demonstrate that the proposed approach is accurate, efficient and robust.